/ 
Paranoid-BerkeleyDB-2.02
River stage zero No dependents
/ Paranoid::BerkeleyDB

NAME

Paranoid::BerkeleyDB -- BerkeleyDB Wrapper

VERSION

$Id: lib/Paranoid/BerkeleyDB.pm, 2.02 2016/06/21 19:51:06 acorliss Exp $

SYNOPSIS

tie %db, Paranoid::BerkeleyDB, 
  Home     => './dbdir',
  Filename => 'data.db';

# Normal hash activities...

# Ensure atomic updates
$dref = tied %db;
$lock = $dref->cds_lock;
$db{$key}++;
$lock->cds_unlock;

untie %db;

DESCRIPTION

This module provides an OO/tie-based wrapper for BerkeleyDB CDS implementations intended for use in tied hashes.

NOTE: This module breaks significantly with previous incarnations of this module. The primary differences are as follows:

Pros
-------------------------------------------------------------
* Places no limitations on the developer regarding BerekelyDB
  environment and database options
* Automatically reuses existing environments for multiple 
  tied hashses
* Uses Btree databases in lieu of hashes, which tended to 
  have issues when the database size grew too large
* Has a fully implemented tied hash interface incorporating 
  CDS locks
* Has pervasive debugging built in using L<Paranoid::Debug>

Cons
-------------------------------------------------------------
* Is no longer considered fork-safe, attempted accesses will
  case the child process to B<croak>.
* Uses Btree databases in lieu of hashes, which does add 
  some additional memory overhead

SUBROUTINES/METHODS

new

tie %db, Paranoid::BerkeleyDB, 
  Home     => './dbdir',
  Filename => 'data.db';

This method is called implicitly when an object is tied. It supports a few differnet invocation styles. The simplest involves simply providing the Home and Filename options. This will set up a CDS environment using the defaults documented in Paranoid::BerkeleyDB::Env(3) and Paranoid::BerkeleyDB::Db(3).

Alternately, you can provide it with Filename and a Paranoid::BerkeleyDB::Env(3) object (or subclassed object) that you instantiated yourself:

tie %db, Paranoid::BerkeleyDB, 
  Env      => $env,
  Filename => 'data.db';

Finally, you can provide it with two hash options to fully control the environment and database instantiation of Paranoid::BerkeleyDB::Env(3) and Paranoid::BerkeleyDB::Db(3):

tie %db, Paranoid::BerkeleyDB, 
  Env      => { %envOpts },
  Db       => { %dbOpts };

dbh

$dref = tied %db;
$dbh  = $dref->dbh;

This method provides access to the BerkeleyDB::Btree(3) object reference.

cds_lock

$dref = tied %db;
$lock = $dref->cds_lock;

This methd provides access to the CDS locks for atomic updates.

DEPENDENCIES

o

BerkeleyDB

o

Carp

o

Class::EHierarchy

o

Fcntl

o

Paranoid

o

Paranoid::BerkeleyDB::Db

o

Paranoid::BerkeleyDB::Env

o

Paranoid::Debug

o

Paranoid::IO

o

Paranoid::IO::Lockfile

BUGS AND LIMITATIONS

-Filename is interpreted differently depending on whether you're using an environment or not. If you're using this module as a standalone DB object any relative paths are interpreted according to your current working directory. If you are using an environment, however, it is interpreted relative to that environment's -Home.

SEE ALSO

L<BerkeleyDB(3)>, L<Paranoid::BerkeleyDB::Env>,
L<Paranoid::BerkeleyDB::Db>

HISTORY

02/12/2016 Complete rewrite

AUTHOR

Arthur Corliss (corliss@digitalmages.com)

LICENSE AND COPYRIGHT

This software is licensed under the same terms as Perl, itself. Please see http://dev.perl.org/licenses/ for more information.

(c) 2005 - 2016, Arthur Corliss (corliss@digitalmages.com)